An unorganized point cloud simplification based on boundary point extraction

In the reverse engineering, the dense and disordered point cloud data contain a huge number of redundancy, which inevitably leads to the significant challenges for the tasks of the subsequent data processing. This paper presents a single axis searching arithmetic to obtain the neighborhood information of a point cloud, and then based on all boundary points extracted and reserved, a non-uniform data reduction scheme, according to a specified curvature threshold and the proportion of reserved points in the k-nearest neighbors, is proposed. The experimental result shows that this approach has a strong ability for identifying boundary points, and can directly and effectively reduce the point cloud data, meanwhile keep the original geometric feature.